stealthstack.ai Status Publish a catalog
Publish

Publish your catalog

StealthStack indexes what publishers list about themselves. Put an ARD catalog on your own domain, check it, then add your domain below. Our crawler reads the files your site publishes and indexes what validates.

  1. Publish an ARD catalog on your domain

    Serve a JSON file at https://your-domain/.well-known/ard.json that lists your skills, MCP servers, agents and APIs. Each entry's identifier names your domain (urn:air:your-domain:…), which is how we tell that the listing came from you. The ARD spec (v0.91) has the full format, and ARD for your business walks through it in plain terms.

    {
  "entries": [
    {
      "identifier": "urn:air:example.com:mcp:invoices",
      "displayName": "Invoices MCP server",
      "description": "Create, search and send invoices.",
      "type": "application/mcp-server-card+json",
      "url": "https://example.com/.well-known/mcp/server-card.json"
    }
  ]
}

  2. Check it

    Run the agent-ready check on your domain. It fetches your catalog the way our crawler does and shows what passes and what to fix.

  3. Add your domain

    We'll read these files from your site, and only these:

    • /.well-known/ard.json ARD catalog: the list of your skills, MCP servers, agents and APIs (ARD v0.91)
    • /.well-known/ai-catalog.json Legacy catalog: the older name for the same catalog
    • /.well-known/agent-card.json A2A agent card: describes one agent
    • /.well-known/mcp/server-card.json MCP server card: describes one MCP server

    A card on its own is indexed too: each becomes one resource, marked unverified because no catalog anchors it. Listing it in your ARD catalog lets it be checked against your domain.

    Just the domain. We only read files at its /.well-known/ paths.

    We keep only the domain, when you sent it and what our crawler found, and we don't publish what you type. To stop abuse we keep a scrambled (keyed hash) copy of your IP address for 2 days. Privacy

What happens next

  • On an upcoming crawl, usually within a day, we fetch the files above from your domain. We follow your robots.txt.
  • If they're valid, the resources they describe are indexed and checked the same way as every other publisher's, including the trust status. From then on your domain is crawled regularly, so updates to your files are picked up.
  • Submitting doesn't guarantee a listing. We skip files that don't validate, and we may leave out sites that list nothing an agent can use.
  • We never publish what you type in this form. Only what your site publishes itself can appear in the index.